* [net-next PATCH 0/3] octeontx2-af: Introduce DMAC based switching
@ 2021-07-19 8:59 Subbaraya Sundeep
2021-07-19 8:59 ` [net-next PATCH 1/3] octeontx2-af: Enable transmit side LBK link Subbaraya Sundeep
` (2 more replies)
0 siblings, 3 replies; 6+ messages in thread
From: Subbaraya Sundeep @ 2021-07-19 8:59 UTC (permalink / raw)
To: davem, kuba, netdev; +Cc: sgoutham, hkelam, gakula, Subbaraya Sundeep
With this patch set packets can be switched between
all CGX mapped PFs and VFs in the system based on
the DMAC addresses. To implement this:
AF allocates high priority rules from top entry(0) in MCAM.
Rules are allocated for all the CGX mapped PFs and VFs though
they are not active and with no NIXLFs attached.
Rules for a PF/VF will be enabled only after they are brought up.
Two rules one for TX and one for RX are allocated for each PF/VF.
A packet sent from a PF/VF with a destination mac of another
PF/VF will be hit by TX rule and sent to LBK channel 63. The
same returned packet will be hit by RX rule whose action is
to forward packet to PF/VF with that destination mac.
Implementation of this for 98xx is tricky since there are
two NIX blocks and till now a PF/VF can install rule for
an NIX0/1 interface only if it is mapped to corresponding NIX0/1 block.
Hence Tx rules are modified such that TX interface in MCAM
entry can be either NIX0-TX or NIX1-TX.
Testing:
1. Create two VFs over PF1(on NIX0) and assign two VFs to two VMs
2. Assign ip addresses to two VFs in VMs and PF2(on NIX1) in host.
3. Assign static arp entries in two VMs and PF2.
4. Ping between VMs and host PF2.
Thanks,
Sundeep
Subbaraya Sundeep (3):
octeontx2-af: Enable transmit side LBK link
octeontx2-af: Prepare for allocating MCAM rules for AF
octeontx2-af: Introduce internal packet switching
drivers/net/ethernet/marvell/octeontx2/af/Makefile | 2 +-
drivers/net/ethernet/marvell/octeontx2/af/rvu.c | 10 +-
drivers/net/ethernet/marvell/octeontx2/af/rvu.h | 21 ++
.../net/ethernet/marvell/octeontx2/af/rvu_cgx.c | 3 +
.../ethernet/marvell/octeontx2/af/rvu_debugfs.c | 5 +-
.../ethernet/marvell/octeontx2/af/rvu_devlink.c | 48 +++-
.../net/ethernet/marvell/octeontx2/af/rvu_nix.c | 36 +++
.../net/ethernet/marvell/octeontx2/af/rvu_npc.c | 47 +++-
.../net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c | 29 ++-
.../net/ethernet/marvell/octeontx2/af/rvu_switch.c | 258 +++++++++++++++++++++
10 files changed, 427 insertions(+), 32 deletions(-)
create mode 100644 drivers/net/ethernet/marvell/octeontx2/af/rvu_switch.c
--
2.7.4
^ permalink raw reply [flat|nested] 6+ messages in thread
* [net-next PATCH 1/3] octeontx2-af: Enable transmit side LBK link
2021-07-19 8:59 [net-next PATCH 0/3] octeontx2-af: Introduce DMAC based switching Subbaraya Sundeep
@ 2021-07-19 8:59 ` Subbaraya Sundeep
2021-07-19 8:59 ` [net-next PATCH 2/3] octeontx2-af: Prepare for allocating MCAM rules for AF Subbaraya Sundeep
2021-07-19 8:59 ` [net-next PATCH 3/3] octeontx2-af: Introduce internal packet switching Subbaraya Sundeep
2 siblings, 0 replies; 6+ messages in thread
From: Subbaraya Sundeep @ 2021-07-19 8:59 UTC (permalink / raw)
To: davem, kuba, netdev; +Cc: sgoutham, hkelam, gakula, Subbaraya Sundeep
For enabling VF-VF switching the packets egressing
out of CGX mapped VFs needed to be sent to LBK
so that same packets are received back to the system.
But the LBK link also needs to be enabled in addition
to a VF's mapped CGX_LMAC link otherwise hardware
raises send error interrupt indicating selected LBK
link is not enabled in NIX_AF_TL3_TL2X_LINKX_CFG register.
Hence this patch enables all LBK links in
TL3_TL2_LINKX_CFG registers.
Also to enable packet flow between PFs/VFs of NIX0
to PFs/VFs of NIX1(in 98xx silicon) the NPC TX DMAC
rules has to be installed such that rules must be hit
for any TX interface i.e., NIX0-TX or NIX1-TX provided
DMAC match creteria is met. Hence this patch changes the
behavior such that MCAM is programmed to match with any
NIX0/1-TX interface for TX rules.
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
---
drivers/net/ethernet/marvell/octeontx2/af/rvu.c | 6 ++++
drivers/net/ethernet/marvell/octeontx2/af/rvu.h | 2 ++
.../net/ethernet/marvell/octeontx2/af/rvu_nix.c | 32 ++++++++++++++++++++++
.../net/ethernet/marvell/octeontx2/af/rvu_npc.c | 15 ++++++++--
.../net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c | 9 +++++-
5 files changed, 61 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
index 10cddf1..086eb6d 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
@@ -2859,6 +2859,12 @@ static int rvu_enable_sriov(struct rvu *rvu)
if (!vfs)
return 0;
+ /* LBK channel number 63 is used for switching packets between
+ * CGX mapped VFs. Hence limit LBK pairs till 62 only.
+ */
+ if (vfs > 62)
+ vfs = 62;
+
/* Save VFs number for reference in VF interrupts handlers.
* Since interrupts might start arriving during SRIOV enablement
* ordinary API cannot be used to get number of enabled VFs.
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
index 10e58a5..e53f530 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
@@ -415,6 +415,8 @@ struct npc_kpu_profile_adapter {
size_t kpus;
};
+#define RVU_SWITCH_LBK_CHAN 63
+
struct rvu {
void __iomem *afreg_base;
void __iomem *pfreg_base;
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
index aeae377..a2d69ea 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
@@ -1952,6 +1952,35 @@ static void nix_tl1_default_cfg(struct rvu *rvu, struct nix_hw *nix_hw,
pfvf_map[schq] = TXSCH_SET_FLAG(pfvf_map[schq], NIX_TXSCHQ_CFG_DONE);
}
+static void rvu_nix_tx_tl2_cfg(struct rvu *rvu, int blkaddr,
+ u16 pcifunc, struct nix_txsch *txsch)
+{
+ struct rvu_hwinfo *hw = rvu->hw;
+ int lbk_link_start, lbk_links;
+ u8 pf = rvu_get_pf(pcifunc);
+ int schq;
+
+ if (!is_pf_cgxmapped(rvu, pf))
+ return;
+
+ lbk_link_start = hw->cgx_links;
+
+ for (schq = 0; schq < txsch->schq.max; schq++) {
+ if (TXSCH_MAP_FUNC(txsch->pfvf_map[schq]) != pcifunc)
+ continue;
+ /* Enable all LBK links with channel 63 by default so that
+ * packets can be sent to LBK with a NPC TX MCAM rule
+ */
+ lbk_links = hw->lbk_links;
+ while (lbk_links--)
+ rvu_write64(rvu, blkaddr,
+ NIX_AF_TL3_TL2X_LINKX_CFG(schq,
+ lbk_link_start +
+ lbk_links),
+ BIT_ULL(12) | RVU_SWITCH_LBK_CHAN);
+ }
+}
+
int rvu_mbox_handler_nix_txschq_cfg(struct rvu *rvu,
struct nix_txschq_config *req,
struct msg_rsp *rsp)
@@ -2040,6 +2069,9 @@ int rvu_mbox_handler_nix_txschq_cfg(struct rvu *rvu,
rvu_write64(rvu, blkaddr, reg, regval);
}
+ rvu_nix_tx_tl2_cfg(rvu, blkaddr, pcifunc,
+ &nix_hw->txsch[NIX_TXSCH_LVL_TL2]);
+
return 0;
}
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
index 3612e0a..16c557c 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
@@ -468,6 +468,8 @@ static void npc_config_mcam_entry(struct rvu *rvu, struct npc_mcam *mcam,
{
int bank = npc_get_bank(mcam, index);
int kw = 0, actbank, actindex;
+ u8 tx_intf_mask = ~intf & 0x3;
+ u8 tx_intf = intf;
u64 cam0, cam1;
actbank = bank; /* Save bank id, to set action later on */
@@ -488,12 +490,21 @@ static void npc_config_mcam_entry(struct rvu *rvu, struct npc_mcam *mcam,
*/
for (; bank < (actbank + mcam->banks_per_entry); bank++, kw = kw + 2) {
/* Interface should be set in all banks */
+ if (is_npc_intf_tx(intf)) {
+ /* Last bit must be set and rest don't care
+ * for TX interfaces
+ */
+ tx_intf_mask = 0x1;
+ tx_intf = intf & tx_intf_mask;
+ tx_intf_mask = ~tx_intf & tx_intf_mask;
+ }
+
rvu_write64(rvu, blkaddr,
NPC_AF_MCAMEX_BANKX_CAMX_INTF(index, bank, 1),
- intf);
+ tx_intf);
rvu_write64(rvu, blkaddr,
NPC_AF_MCAMEX_BANKX_CAMX_INTF(index, bank, 0),
- ~intf & 0x3);
+ tx_intf_mask);
/* Set the match key */
npc_get_keyword(entry, kw, &cam0, &cam1);
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
index 6863314..92d64bd 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
@@ -949,9 +949,16 @@ static void npc_update_tx_entry(struct rvu *rvu, struct rvu_pfvf *pfvf,
struct npc_install_flow_req *req, u16 target)
{
struct nix_tx_action action;
+ u64 mask = ~0ULL;
+
+ /* If AF is installing then do not care about
+ * PF_FUNC in Send Descriptor
+ */
+ if (is_pffunc_af(req->hdr.pcifunc))
+ mask = 0;
npc_update_entry(rvu, NPC_PF_FUNC, entry, (__force u16)htons(target),
- 0, ~0ULL, 0, NIX_INTF_TX);
+ 0, mask, 0, NIX_INTF_TX);
*(u64 *)&action = 0x00;
action.op = req->op;
--
2.7.4
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [net-next PATCH 2/3] octeontx2-af: Prepare for allocating MCAM rules for AF
2021-07-19 8:59 [net-next PATCH 0/3] octeontx2-af: Introduce DMAC based switching Subbaraya Sundeep
2021-07-19 8:59 ` [net-next PATCH 1/3] octeontx2-af: Enable transmit side LBK link Subbaraya Sundeep
@ 2021-07-19 8:59 ` Subbaraya Sundeep
2021-07-19 8:59 ` [net-next PATCH 3/3] octeontx2-af: Introduce internal packet switching Subbaraya Sundeep
2 siblings, 0 replies; 6+ messages in thread
From: Subbaraya Sundeep @ 2021-07-19 8:59 UTC (permalink / raw)
To: davem, kuba, netdev; +Cc: sgoutham, hkelam, gakula, Subbaraya Sundeep
AF till now only manages the allocation and freeing of
MCAM rules for other PF/VFs in system. To implement
L2 switching between all CGX mapped PF and VFs, AF
requires MCAM entries for DMAC rules for each PF and VF.
This patch modifies AF driver such that AF can also
allocate MCAM rules and install rules for other
PFs and VFs. All the checks like channel verification
for RX rules and PF_FUNC verification for TX rules are
relaxed in case AF is allocating or installing rules.
Also all the entry and counter to owner mappings are
set to NPC_MCAM_INVALID_MAP when they are free indicating
those are not allocated to AF nor PF/VFs.
This patch also ensures that AF allocated and installed
entries are displayed in debugfs.
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
---
.../ethernet/marvell/octeontx2/af/rvu_debugfs.c | 5 +---
.../net/ethernet/marvell/octeontx2/af/rvu_npc.c | 32 ++++++++++++++++------
.../net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c | 11 ++++----
3 files changed, 30 insertions(+), 18 deletions(-)
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
index 370d4ca..9b2dfbf 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
@@ -2113,9 +2113,6 @@ static void rvu_print_npc_mcam_info(struct seq_file *s,
int entry_acnt, entry_ecnt;
int cntr_acnt, cntr_ecnt;
- /* Skip PF0 */
- if (!pcifunc)
- return;
rvu_npc_get_mcam_entry_alloc_info(rvu, pcifunc, blkaddr,
&entry_acnt, &entry_ecnt);
rvu_npc_get_mcam_counter_alloc_info(rvu, pcifunc, blkaddr,
@@ -2298,7 +2295,7 @@ static void rvu_dbg_npc_mcam_show_flows(struct seq_file *s,
static void rvu_dbg_npc_mcam_show_action(struct seq_file *s,
struct rvu_npc_mcam_rule *rule)
{
- if (rule->intf == NIX_INTF_TX) {
+ if (is_npc_intf_tx(rule->intf)) {
switch (rule->tx_action.op) {
case NIX_TX_ACTIONOP_DROP:
seq_puts(s, "\taction: Drop\n");
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
index 16c557c..1097291 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
@@ -442,7 +442,8 @@ static void npc_fixup_vf_rule(struct rvu *rvu, struct npc_mcam *mcam,
owner = mcam->entry2pfvf_map[index];
target_func = (entry->action >> 4) & 0xffff;
/* do nothing when target is LBK/PF or owner is not PF */
- if (is_afvf(target_func) || (owner & RVU_PFVF_FUNC_MASK) ||
+ if (is_pffunc_af(owner) || is_afvf(target_func) ||
+ (owner & RVU_PFVF_FUNC_MASK) ||
!(target_func & RVU_PFVF_FUNC_MASK))
return;
@@ -661,6 +662,7 @@ void rvu_npc_install_ucast_entry(struct rvu *rvu, u16 pcifunc,
eth_broadcast_addr((u8 *)&req.mask.dmac);
req.features = BIT_ULL(NPC_DMAC);
req.channel = chan;
+ req.chan_mask = 0xFFFU;
req.intf = pfvf->nix_rx_intf;
req.op = action.op;
req.hdr.pcifunc = 0; /* AF is requester */
@@ -810,6 +812,7 @@ void rvu_npc_install_bcast_match_entry(struct rvu *rvu, u16 pcifunc,
eth_broadcast_addr((u8 *)&req.mask.dmac);
req.features = BIT_ULL(NPC_DMAC);
req.channel = chan;
+ req.chan_mask = 0xFFFU;
req.intf = pfvf->nix_rx_intf;
req.entry = index;
req.hdr.pcifunc = 0; /* AF is requester */
@@ -1756,6 +1759,8 @@ static int npc_mcam_rsrcs_init(struct rvu *rvu, int blkaddr)
int nixlf_count = rvu_get_nixlf_count(rvu);
struct npc_mcam *mcam = &rvu->hw->mcam;
int rsvd, err;
+ u16 index;
+ int cntr;
u64 cfg;
/* Actual number of MCAM entries vary by entry size */
@@ -1856,6 +1861,14 @@ static int npc_mcam_rsrcs_init(struct rvu *rvu, int blkaddr)
if (!mcam->entry2target_pffunc)
goto free_mem;
+ for (index = 0; index < mcam->bmap_entries; index++) {
+ mcam->entry2pfvf_map[index] = NPC_MCAM_INVALID_MAP;
+ mcam->entry2cntr_map[index] = NPC_MCAM_INVALID_MAP;
+ }
+
+ for (cntr = 0; cntr < mcam->counters.max; cntr++)
+ mcam->cntr2pfvf_map[cntr] = NPC_MCAM_INVALID_MAP;
+
mutex_init(&mcam->lock);
return 0;
@@ -2573,7 +2586,7 @@ int rvu_mbox_handler_npc_mcam_alloc_entry(struct rvu *rvu,
}
/* Alloc request from PFFUNC with no NIXLF attached should be denied */
- if (!is_nixlf_attached(rvu, pcifunc))
+ if (!is_pffunc_af(pcifunc) && !is_nixlf_attached(rvu, pcifunc))
return NPC_MCAM_ALLOC_DENIED;
return npc_mcam_alloc_entries(mcam, pcifunc, req, rsp);
@@ -2593,7 +2606,7 @@ int rvu_mbox_handler_npc_mcam_free_entry(struct rvu *rvu,
return NPC_MCAM_INVALID_REQ;
/* Free request from PFFUNC with no NIXLF attached, ignore */
- if (!is_nixlf_attached(rvu, pcifunc))
+ if (!is_pffunc_af(pcifunc) && !is_nixlf_attached(rvu, pcifunc))
return NPC_MCAM_INVALID_REQ;
mutex_lock(&mcam->lock);
@@ -2605,7 +2618,7 @@ int rvu_mbox_handler_npc_mcam_free_entry(struct rvu *rvu,
if (rc)
goto exit;
- mcam->entry2pfvf_map[req->entry] = 0;
+ mcam->entry2pfvf_map[req->entry] = NPC_MCAM_INVALID_MAP;
mcam->entry2target_pffunc[req->entry] = 0x0;
npc_mcam_clear_bit(mcam, req->entry);
npc_enable_mcam_entry(rvu, mcam, blkaddr, req->entry, false);
@@ -2690,13 +2703,14 @@ int rvu_mbox_handler_npc_mcam_write_entry(struct rvu *rvu,
else
nix_intf = pfvf->nix_rx_intf;
- if (npc_mcam_verify_channel(rvu, pcifunc, req->intf, channel)) {
+ if (!is_pffunc_af(pcifunc) &&
+ npc_mcam_verify_channel(rvu, pcifunc, req->intf, channel)) {
rc = NPC_MCAM_INVALID_REQ;
goto exit;
}
- if (npc_mcam_verify_pf_func(rvu, &req->entry_data, req->intf,
- pcifunc)) {
+ if (!is_pffunc_af(pcifunc) &&
+ npc_mcam_verify_pf_func(rvu, &req->entry_data, req->intf, pcifunc)) {
rc = NPC_MCAM_INVALID_REQ;
goto exit;
}
@@ -2847,7 +2861,7 @@ int rvu_mbox_handler_npc_mcam_alloc_counter(struct rvu *rvu,
return NPC_MCAM_INVALID_REQ;
/* If the request is from a PFFUNC with no NIXLF attached, ignore */
- if (!is_nixlf_attached(rvu, pcifunc))
+ if (!is_pffunc_af(pcifunc) && !is_nixlf_attached(rvu, pcifunc))
return NPC_MCAM_INVALID_REQ;
/* Since list of allocated counter IDs needs to be sent to requester,
@@ -3092,7 +3106,7 @@ int rvu_mbox_handler_npc_mcam_alloc_and_write_entry(struct rvu *rvu,
if (rc) {
/* Free allocated MCAM entry */
mutex_lock(&mcam->lock);
- mcam->entry2pfvf_map[entry] = 0;
+ mcam->entry2pfvf_map[entry] = NPC_MCAM_INVALID_MAP;
npc_mcam_clear_bit(mcam, entry);
mutex_unlock(&mcam->lock);
return rc;
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
index 92d64bd..c1f35a0 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
@@ -913,11 +913,9 @@ static void npc_update_rx_entry(struct rvu *rvu, struct rvu_pfvf *pfvf,
struct npc_install_flow_req *req, u16 target)
{
struct nix_rx_action action;
- u64 chan_mask;
- chan_mask = req->chan_mask ? req->chan_mask : ~0ULL;
- npc_update_entry(rvu, NPC_CHAN, entry, req->channel, 0, chan_mask, 0,
- NIX_INTF_RX);
+ npc_update_entry(rvu, NPC_CHAN, entry, req->channel, 0, req->chan_mask,
+ 0, NIX_INTF_RX);
*(u64 *)&action = 0x00;
action.pf_func = target;
@@ -1171,7 +1169,9 @@ int rvu_mbox_handler_npc_install_flow(struct rvu *rvu,
if (err)
return err;
- if (npc_mcam_verify_channel(rvu, target, req->intf, req->channel))
+ /* Skip channel validation if AF is installing */
+ if (!is_pffunc_af(req->hdr.pcifunc) &&
+ npc_mcam_verify_channel(rvu, target, req->intf, req->channel))
return -EINVAL;
pfvf = rvu_get_pfvf(rvu, target);
@@ -1187,6 +1187,7 @@ int rvu_mbox_handler_npc_install_flow(struct rvu *rvu,
eth_broadcast_addr((u8 *)&req->mask.dmac);
}
+ /* Proceed if NIXLF is attached or not for TX rules */
err = nix_get_nixlf(rvu, target, &nixlf, NULL);
if (err && is_npc_intf_rx(req->intf) && !pf_set_vfs_mac)
return -EINVAL;
--
2.7.4
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [net-next PATCH 3/3] octeontx2-af: Introduce internal packet switching
2021-07-19 8:59 [net-next PATCH 0/3] octeontx2-af: Introduce DMAC based switching Subbaraya Sundeep
2021-07-19 8:59 ` [net-next PATCH 1/3] octeontx2-af: Enable transmit side LBK link Subbaraya Sundeep
2021-07-19 8:59 ` [net-next PATCH 2/3] octeontx2-af: Prepare for allocating MCAM rules for AF Subbaraya Sundeep
@ 2021-07-19 8:59 ` Subbaraya Sundeep
2021-07-20 11:51 ` Jakub Kicinski
2 siblings, 1 reply; 6+ messages in thread
From: Subbaraya Sundeep @ 2021-07-19 8:59 UTC (permalink / raw)
To: davem, kuba, netdev; +Cc: sgoutham, hkelam, gakula, Subbaraya Sundeep
As of now any communication between CGXs PFs and
their VFs within the system is possible only by
external switches sending packets back to the
system. This patch adds internal switching support.
Broadcast packet replication is not covered here.
RVU admin function (AF) maintains MAC addresses
of all interfaces in the system. When switching is
enabled, MCAM entries are allocated to install rules
such that packets with DMAC matching any of the
internal interface MAC addresses is punted back
into the system via the loopback channel.
On the receive side the default unicast rules
are modified to not check for ingress channel.
So any packet with matching DMAC irrespective of
which interface it is coming from will be forwarded
to the respective PF/VF interface.
The transmit side rules and default unicast rules
are updated if user changes MAC address of an interface.
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
---
drivers/net/ethernet/marvell/octeontx2/af/Makefile | 2 +-
drivers/net/ethernet/marvell/octeontx2/af/rvu.c | 4 +-
drivers/net/ethernet/marvell/octeontx2/af/rvu.h | 19 ++
.../net/ethernet/marvell/octeontx2/af/rvu_cgx.c | 3 +
.../ethernet/marvell/octeontx2/af/rvu_devlink.c | 48 +++-
.../net/ethernet/marvell/octeontx2/af/rvu_nix.c | 4 +
.../net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c | 9 +-
.../net/ethernet/marvell/octeontx2/af/rvu_switch.c | 258 +++++++++++++++++++++
8 files changed, 336 insertions(+), 11 deletions(-)
create mode 100644 drivers/net/ethernet/marvell/octeontx2/af/rvu_switch.c
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/Makefile b/drivers/net/ethernet/marvell/octeontx2/af/Makefile
index 1a34556..cc8ac36 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/Makefile
+++ b/drivers/net/ethernet/marvell/octeontx2/af/Makefile
@@ -10,4 +10,4 @@ obj-$(CONFIG_OCTEONTX2_AF) += rvu_af.o
rvu_mbox-y := mbox.o rvu_trace.o
rvu_af-y := cgx.o rvu.o rvu_cgx.o rvu_npa.o rvu_nix.o \
rvu_reg.o rvu_npc.o rvu_debugfs.o ptp.o rvu_npc_fs.o \
- rvu_cpt.o rvu_devlink.o rpm.o rvu_cn10k.o
+ rvu_cpt.o rvu_devlink.o rpm.o rvu_cn10k.o rvu_switch.o
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
index 086eb6d..017163f 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
@@ -1314,7 +1314,7 @@ int rvu_mbox_handler_detach_resources(struct rvu *rvu,
return rvu_detach_rsrcs(rvu, detach, detach->hdr.pcifunc);
}
-static int rvu_get_nix_blkaddr(struct rvu *rvu, u16 pcifunc)
+int rvu_get_nix_blkaddr(struct rvu *rvu, u16 pcifunc)
{
struct rvu_pfvf *pfvf = rvu_get_pfvf(rvu, pcifunc);
int blkaddr = BLKADDR_NIX0, vf;
@@ -3007,6 +3007,8 @@ static int rvu_probe(struct pci_dev *pdev, const struct pci_device_id *id)
/* Initialize debugfs */
rvu_dbg_init(rvu);
+ mutex_init(&rvu->rswitch.switch_lock);
+
return 0;
err_dl:
rvu_unregister_dl(rvu);
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
index e53f530..91503fb 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
@@ -417,6 +417,14 @@ struct npc_kpu_profile_adapter {
#define RVU_SWITCH_LBK_CHAN 63
+struct rvu_switch {
+ struct mutex switch_lock; /* Serialize flow installation */
+ u32 used_entries;
+ u16 *entry2pcifunc;
+ u16 mode;
+ u16 start_entry;
+};
+
struct rvu {
void __iomem *afreg_base;
void __iomem *pfreg_base;
@@ -447,6 +455,7 @@ struct rvu {
/* CGX */
#define PF_CGXMAP_BASE 1 /* PF 0 is reserved for RVU PF */
+ u16 cgx_mapped_vfs; /* maximum CGX mapped VFs */
u8 cgx_mapped_pfs;
u8 cgx_cnt_max; /* CGX port count max */
u8 *pf2cgxlmac_map; /* pf to cgx_lmac map */
@@ -479,6 +488,9 @@ struct rvu {
struct rvu_debugfs rvu_dbg;
#endif
struct rvu_devlink *rvu_dl;
+
+ /* RVU switch implementation over NPC with DMAC rules */
+ struct rvu_switch rswitch;
};
static inline void rvu_write64(struct rvu *rvu, u64 block, u64 offset, u64 val)
@@ -693,6 +705,7 @@ int nix_aq_context_read(struct rvu *rvu, struct nix_hw *nix_hw,
struct nix_cn10k_aq_enq_req *aq_req,
struct nix_cn10k_aq_enq_rsp *aq_rsp,
u16 pcifunc, u8 ctype, u32 qidx);
+int rvu_get_nix_blkaddr(struct rvu *rvu, u16 pcifunc);
/* NPC APIs */
int rvu_npc_init(struct rvu *rvu);
@@ -770,4 +783,10 @@ void rvu_dbg_exit(struct rvu *rvu);
static inline void rvu_dbg_init(struct rvu *rvu) {}
static inline void rvu_dbg_exit(struct rvu *rvu) {}
#endif
+
+/* RVU Switch */
+void rvu_switch_enable(struct rvu *rvu);
+void rvu_switch_disable(struct rvu *rvu);
+void rvu_switch_update_rules(struct rvu *rvu, u16 pcifunc);
+
#endif /* RVU_H */
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
index 6cc8fbb..fe99ac4 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
@@ -126,6 +126,7 @@ static int rvu_map_cgx_lmac_pf(struct rvu *rvu)
unsigned long lmac_bmap;
int size, free_pkind;
int cgx, lmac, iter;
+ int numvfs, hwvfs;
if (!cgx_cnt_max)
return 0;
@@ -166,6 +167,8 @@ static int rvu_map_cgx_lmac_pf(struct rvu *rvu)
pkind->pfchan_map[free_pkind] = ((pf) & 0x3F) << 16;
rvu_map_cgx_nix_block(rvu, pf, cgx, lmac);
rvu->cgx_mapped_pfs++;
+ rvu_get_pf_numvfs(rvu, pf, &numvfs, &hwvfs);
+ rvu->cgx_mapped_vfs += numvfs;
pf++;
}
}
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_devlink.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_devlink.c
index 10a98bc..2688186 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_devlink.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_devlink.c
@@ -1364,6 +1364,44 @@ static void rvu_health_reporters_destroy(struct rvu *rvu)
rvu_nix_health_reporters_destroy(rvu_dl);
}
+static int rvu_devlink_eswitch_mode_get(struct devlink *devlink, u16 *mode)
+{
+ struct rvu_devlink *rvu_dl = devlink_priv(devlink);
+ struct rvu *rvu = rvu_dl->rvu;
+ struct rvu_switch *rswitch;
+
+ rswitch = &rvu->rswitch;
+ *mode = rswitch->mode;
+
+ return 0;
+}
+
+static int rvu_devlink_eswitch_mode_set(struct devlink *devlink, u16 mode,
+ struct netlink_ext_ack *extack)
+{
+ struct rvu_devlink *rvu_dl = devlink_priv(devlink);
+ struct rvu *rvu = rvu_dl->rvu;
+ struct rvu_switch *rswitch;
+
+ rswitch = &rvu->rswitch;
+ switch (mode) {
+ case DEVLINK_ESWITCH_MODE_LEGACY:
+ case DEVLINK_ESWITCH_MODE_SWITCHDEV:
+ if (rswitch->mode == mode)
+ return 0;
+ rswitch->mode = mode;
+ if (mode == DEVLINK_ESWITCH_MODE_SWITCHDEV)
+ rvu_switch_enable(rvu);
+ else
+ rvu_switch_disable(rvu);
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
static int rvu_devlink_info_get(struct devlink *devlink, struct devlink_info_req *req,
struct netlink_ext_ack *extack)
{
@@ -1372,6 +1410,8 @@ static int rvu_devlink_info_get(struct devlink *devlink, struct devlink_info_req
static const struct devlink_ops rvu_devlink_ops = {
.info_get = rvu_devlink_info_get,
+ .eswitch_mode_get = rvu_devlink_eswitch_mode_get,
+ .eswitch_mode_set = rvu_devlink_eswitch_mode_set,
};
int rvu_register_dl(struct rvu *rvu)
@@ -1380,14 +1420,9 @@ int rvu_register_dl(struct rvu *rvu)
struct devlink *dl;
int err;
- rvu_dl = kzalloc(sizeof(*rvu_dl), GFP_KERNEL);
- if (!rvu_dl)
- return -ENOMEM;
-
dl = devlink_alloc(&rvu_devlink_ops, sizeof(struct rvu_devlink));
if (!dl) {
dev_warn(rvu->dev, "devlink_alloc failed\n");
- kfree(rvu_dl);
return -ENOMEM;
}
@@ -1395,10 +1430,10 @@ int rvu_register_dl(struct rvu *rvu)
if (err) {
dev_err(rvu->dev, "devlink register failed with error %d\n", err);
devlink_free(dl);
- kfree(rvu_dl);
return err;
}
+ rvu_dl = devlink_priv(dl);
rvu_dl->dl = dl;
rvu_dl->rvu = rvu;
rvu->rvu_dl = rvu_dl;
@@ -1417,5 +1452,4 @@ void rvu_unregister_dl(struct rvu *rvu)
rvu_health_reporters_destroy(rvu);
devlink_unregister(dl);
devlink_free(dl);
- kfree(rvu_dl);
}
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
index a2d69ea..0933699 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
@@ -3212,6 +3212,8 @@ int rvu_mbox_handler_nix_set_mac_addr(struct rvu *rvu,
if (test_bit(PF_SET_VF_TRUSTED, &pfvf->flags) && from_vf)
ether_addr_copy(pfvf->default_mac, req->mac_addr);
+ rvu_switch_update_rules(rvu, pcifunc);
+
return 0;
}
@@ -3881,6 +3883,8 @@ int rvu_mbox_handler_nix_lf_start_rx(struct rvu *rvu, struct msg_req *req,
pfvf = rvu_get_pfvf(rvu, pcifunc);
set_bit(NIXLF_INITIALIZED, &pfvf->flags);
+ rvu_switch_update_rules(rvu, pcifunc);
+
return rvu_cgx_start_stop_io(rvu, pcifunc, true);
}
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
index c1f35a0..5c01cf4 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
@@ -910,10 +910,15 @@ static void rvu_mcam_add_counter_to_rule(struct rvu *rvu, u16 pcifunc,
static void npc_update_rx_entry(struct rvu *rvu, struct rvu_pfvf *pfvf,
struct mcam_entry *entry,
- struct npc_install_flow_req *req, u16 target)
+ struct npc_install_flow_req *req,
+ u16 target, bool pf_set_vfs_mac)
{
+ struct rvu_switch *rswitch = &rvu->rswitch;
struct nix_rx_action action;
+ if (rswitch->mode == DEVLINK_ESWITCH_MODE_SWITCHDEV && pf_set_vfs_mac)
+ req->chan_mask = 0x0; /* Do not care channel */
+
npc_update_entry(rvu, NPC_CHAN, entry, req->channel, 0, req->chan_mask,
0, NIX_INTF_RX);
@@ -1007,7 +1012,7 @@ static int npc_install_flow(struct rvu *rvu, int blkaddr, u16 target,
req->intf);
if (is_npc_intf_rx(req->intf))
- npc_update_rx_entry(rvu, pfvf, entry, req, target);
+ npc_update_rx_entry(rvu, pfvf, entry, req, target, pf_set_vfs_mac);
else
npc_update_tx_entry(rvu, pfvf, entry, req, target);
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_switch.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_switch.c
new file mode 100644
index 0000000..2e53797
--- /dev/null
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_switch.c
@@ -0,0 +1,258 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Marvell OcteonTx2 RVU Admin Function driver
+ *
+ * Copyright (C) 2021 Marvell.
+ */
+
+#include <linux/bitfield.h>
+#include "rvu.h"
+
+static int rvu_switch_install_rx_rule(struct rvu *rvu, u16 pcifunc,
+ u16 chan_mask)
+{
+ struct npc_install_flow_req req = { 0 };
+ struct npc_install_flow_rsp rsp = { 0 };
+ struct rvu_pfvf *pfvf;
+
+ pfvf = rvu_get_pfvf(rvu, pcifunc);
+ /* If the pcifunc is not initialized then nothing to do.
+ * This same function will be called again via rvu_switch_update_rules
+ * after pcifunc is initialized.
+ */
+ if (!test_bit(NIXLF_INITIALIZED, &pfvf->flags))
+ return 0;
+
+ ether_addr_copy(req.packet.dmac, pfvf->mac_addr);
+ eth_broadcast_addr((u8 *)&req.mask.dmac);
+ req.hdr.pcifunc = 0; /* AF is requester */
+ req.vf = pcifunc;
+ req.features = BIT_ULL(NPC_DMAC);
+ req.channel = pfvf->rx_chan_base;
+ req.chan_mask = chan_mask;
+ req.intf = pfvf->nix_rx_intf;
+ req.op = NIX_RX_ACTION_DEFAULT;
+ req.default_rule = 1;
+
+ return rvu_mbox_handler_npc_install_flow(rvu, &req, &rsp);
+}
+
+static int rvu_switch_install_tx_rule(struct rvu *rvu, u16 pcifunc, u16 entry)
+{
+ struct npc_install_flow_req req = { 0 };
+ struct npc_install_flow_rsp rsp = { 0 };
+ struct rvu_pfvf *pfvf;
+ u8 lbkid;
+
+ pfvf = rvu_get_pfvf(rvu, pcifunc);
+ /* If the pcifunc is not initialized then nothing to do.
+ * This same function will be called again via rvu_switch_update_rules
+ * after pcifunc is initialized.
+ */
+ if (!test_bit(NIXLF_INITIALIZED, &pfvf->flags))
+ return 0;
+
+ lbkid = pfvf->nix_blkaddr == BLKADDR_NIX0 ? 0 : 1;
+ ether_addr_copy(req.packet.dmac, pfvf->mac_addr);
+ eth_broadcast_addr((u8 *)&req.mask.dmac);
+ req.hdr.pcifunc = 0; /* AF is requester */
+ req.vf = pcifunc;
+ req.entry = entry;
+ req.features = BIT_ULL(NPC_DMAC);
+ req.intf = pfvf->nix_tx_intf;
+ req.op = NIX_TX_ACTIONOP_UCAST_CHAN;
+ req.index = (lbkid << 8) | RVU_SWITCH_LBK_CHAN;
+ req.set_cntr = 1;
+
+ return rvu_mbox_handler_npc_install_flow(rvu, &req, &rsp);
+}
+
+static int rvu_switch_install_rules(struct rvu *rvu)
+{
+ struct rvu_switch *rswitch = &rvu->rswitch;
+ u16 start = rswitch->start_entry;
+ struct rvu_hwinfo *hw = rvu->hw;
+ int pf, vf, numvfs, hwvf;
+ u16 pcifunc, entry = 0;
+ int err;
+
+ for (pf = 1; pf < hw->total_pfs; pf++) {
+ if (!is_pf_cgxmapped(rvu, pf))
+ continue;
+
+ pcifunc = pf << 10;
+ /* rvu_get_nix_blkaddr sets up the corresponding NIX block
+ * address and NIX RX and TX interfaces for a pcifunc.
+ * Generally it is called during attach call of a pcifunc but it
+ * is called here since we are pre-installing rules before
+ * nixlfs are attached
+ */
+ rvu_get_nix_blkaddr(rvu, pcifunc);
+
+ /* MCAM RX rule for a PF/VF already exists as default unicast
+ * rules installed by AF. Hence change the channel in those
+ * rules to ignore channel so that packets with the required
+ * DMAC received from LBK(by other PF/VFs in system) or from
+ * external world (from wire) are accepted.
+ */
+ err = rvu_switch_install_rx_rule(rvu, pcifunc, 0x0);
+ if (err) {
+ dev_err(rvu->dev, "RX rule for PF%d failed(%d)\n",
+ pf, err);
+ return err;
+ }
+
+ err = rvu_switch_install_tx_rule(rvu, pcifunc, start + entry);
+ if (err) {
+ dev_err(rvu->dev, "TX rule for PF%d failed(%d)\n",
+ pf, err);
+ return err;
+ }
+
+ rswitch->entry2pcifunc[entry++] = pcifunc;
+
+ rvu_get_pf_numvfs(rvu, pf, &numvfs, &hwvf);
+ for (vf = 0; vf < numvfs; vf++, hwvf++) {
+ pcifunc = pf << 10 | ((vf + 1) & 0x3FF);
+ rvu_get_nix_blkaddr(rvu, pcifunc);
+
+ err = rvu_switch_install_rx_rule(rvu, pcifunc, 0x0);
+ if (err) {
+ dev_err(rvu->dev,
+ "RX rule for PF%dVF%d failed(%d)\n",
+ pf, vf, err);
+ return err;
+ }
+
+ err = rvu_switch_install_tx_rule(rvu, pcifunc,
+ start + entry);
+ if (err) {
+ dev_err(rvu->dev,
+ "TX rule for PF%dVF%d failed(%d)\n",
+ pf, vf, err);
+ return err;
+ }
+
+ rswitch->entry2pcifunc[entry++] = pcifunc;
+ }
+ }
+
+ return 0;
+}
+
+void rvu_switch_enable(struct rvu *rvu)
+{
+ struct npc_mcam_alloc_entry_req alloc_req = { 0 };
+ struct npc_mcam_alloc_entry_rsp alloc_rsp = { 0 };
+ struct npc_delete_flow_req uninstall_req = { 0 };
+ struct npc_mcam_free_entry_req free_req = { 0 };
+ struct rvu_switch *rswitch = &rvu->rswitch;
+ struct msg_rsp rsp;
+ int ret;
+
+ alloc_req.contig = true;
+ alloc_req.count = rvu->cgx_mapped_pfs + rvu->cgx_mapped_vfs;
+ ret = rvu_mbox_handler_npc_mcam_alloc_entry(rvu, &alloc_req,
+ &alloc_rsp);
+ if (ret) {
+ dev_err(rvu->dev,
+ "Unable to allocate MCAM entries\n");
+ goto exit;
+ }
+
+ if (alloc_rsp.count != alloc_req.count) {
+ dev_err(rvu->dev,
+ "Unable to allocate %d MCAM entries, got %d\n",
+ alloc_req.count, alloc_rsp.count);
+ goto free_entries;
+ }
+
+ rswitch->entry2pcifunc = kcalloc(alloc_req.count, sizeof(u16),
+ GFP_KERNEL);
+ if (!rswitch->entry2pcifunc)
+ goto free_entries;
+
+ rswitch->used_entries = alloc_rsp.count;
+ rswitch->start_entry = alloc_rsp.entry;
+
+ ret = rvu_switch_install_rules(rvu);
+ if (ret)
+ goto uninstall_rules;
+
+ return;
+
+uninstall_rules:
+ uninstall_req.start = rswitch->start_entry;
+ uninstall_req.end = rswitch->start_entry + rswitch->used_entries - 1;
+ rvu_mbox_handler_npc_delete_flow(rvu, &uninstall_req, &rsp);
+ kfree(rswitch->entry2pcifunc);
+free_entries:
+ free_req.all = 1;
+ rvu_mbox_handler_npc_mcam_free_entry(rvu, &free_req, &rsp);
+exit:
+ return;
+}
+
+void rvu_switch_disable(struct rvu *rvu)
+{
+ struct npc_delete_flow_req uninstall_req = { 0 };
+ struct npc_mcam_free_entry_req free_req = { 0 };
+ struct rvu_switch *rswitch = &rvu->rswitch;
+ struct rvu_hwinfo *hw = rvu->hw;
+ int pf, vf, numvfs, hwvf;
+ struct msg_rsp rsp;
+ u16 pcifunc;
+ int err;
+
+ if (!rswitch->used_entries)
+ return;
+
+ for (pf = 1; pf < hw->total_pfs; pf++) {
+ if (!is_pf_cgxmapped(rvu, pf))
+ continue;
+
+ pcifunc = pf << 10;
+ err = rvu_switch_install_rx_rule(rvu, pcifunc, 0xFFF);
+ if (err)
+ dev_err(rvu->dev,
+ "Reverting RX rule for PF%d failed(%d)\n",
+ pf, err);
+
+ for (vf = 0; vf < numvfs; vf++, hwvf++) {
+ pcifunc = pf << 10 | ((vf + 1) & 0x3FF);
+ err = rvu_switch_install_rx_rule(rvu, pcifunc, 0xFFF);
+ if (err)
+ dev_err(rvu->dev,
+ "Reverting RX rule for PF%dVF%d failed(%d)\n",
+ pf, vf, err);
+ }
+ }
+
+ uninstall_req.start = rswitch->start_entry;
+ uninstall_req.end = rswitch->start_entry + rswitch->used_entries - 1;
+ free_req.all = 1;
+ rvu_mbox_handler_npc_delete_flow(rvu, &uninstall_req, &rsp);
+ rvu_mbox_handler_npc_mcam_free_entry(rvu, &free_req, &rsp);
+ rswitch->used_entries = 0;
+ kfree(rswitch->entry2pcifunc);
+}
+
+void rvu_switch_update_rules(struct rvu *rvu, u16 pcifunc)
+{
+ struct rvu_switch *rswitch = &rvu->rswitch;
+ u32 max = rswitch->used_entries;
+ u16 entry;
+
+ if (!rswitch->used_entries)
+ return;
+
+ for (entry = 0; entry < max; entry++) {
+ if (rswitch->entry2pcifunc[entry] == pcifunc)
+ break;
+ }
+
+ if (entry >= max)
+ return;
+
+ rvu_switch_install_tx_rule(rvu, pcifunc, rswitch->start_entry + entry);
+ rvu_switch_install_rx_rule(rvu, pcifunc, 0x0);
+}
--
2.7.4
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [net-next PATCH 3/3] octeontx2-af: Introduce internal packet switching
2021-07-19 8:59 ` [net-next PATCH 3/3] octeontx2-af: Introduce internal packet switching Subbaraya Sundeep
@ 2021-07-20 11:51 ` Jakub Kicinski
2021-07-20 17:56 ` sundeep subbaraya
0 siblings, 1 reply; 6+ messages in thread
From: Jakub Kicinski @ 2021-07-20 11:51 UTC (permalink / raw)
To: Subbaraya Sundeep; +Cc: davem, netdev, sgoutham, hkelam, gakula
On Mon, 19 Jul 2021 14:29:34 +0530, Subbaraya Sundeep wrote:
> +static int rvu_devlink_eswitch_mode_set(struct devlink *devlink, u16 mode,
> + struct netlink_ext_ack *extack)
> +{
> + struct rvu_devlink *rvu_dl = devlink_priv(devlink);
> + struct rvu *rvu = rvu_dl->rvu;
> + struct rvu_switch *rswitch;
> +
> + rswitch = &rvu->rswitch;
> + switch (mode) {
> + case DEVLINK_ESWITCH_MODE_LEGACY:
> + case DEVLINK_ESWITCH_MODE_SWITCHDEV:
> + if (rswitch->mode == mode)
> + return 0;
> + rswitch->mode = mode;
> + if (mode == DEVLINK_ESWITCH_MODE_SWITCHDEV)
> + rvu_switch_enable(rvu);
> + else
> + rvu_switch_disable(rvu);
I don't see the code handle creation and tearing down of representors.
How do things work in this driver? Does AF have a representor netdev
for each VF (that's separate from the VF netdev itself)? Those should
only exist in switchdev mode, while legacy mode should use DMAC
switching.
I think what you want is a textbook VEPA vs VEB switch. Please take a
look at drivers implementing .ndo_bridge_getlink/.ndo_bridge_setlink.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [net-next PATCH 3/3] octeontx2-af: Introduce internal packet switching
2021-07-20 11:51 ` Jakub Kicinski
@ 2021-07-20 17:56 ` sundeep subbaraya
0 siblings, 0 replies; 6+ messages in thread
From: sundeep subbaraya @ 2021-07-20 17:56 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Subbaraya Sundeep, David Miller, netdev, Sunil Kovvuri Goutham,
hariprasad, Geetha sowjanya
Hi Jakub,
On Tue, Jul 20, 2021 at 5:26 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Mon, 19 Jul 2021 14:29:34 +0530, Subbaraya Sundeep wrote:
> > +static int rvu_devlink_eswitch_mode_set(struct devlink *devlink, u16 mode,
> > + struct netlink_ext_ack *extack)
> > +{
> > + struct rvu_devlink *rvu_dl = devlink_priv(devlink);
> > + struct rvu *rvu = rvu_dl->rvu;
> > + struct rvu_switch *rswitch;
> > +
> > + rswitch = &rvu->rswitch;
> > + switch (mode) {
> > + case DEVLINK_ESWITCH_MODE_LEGACY:
> > + case DEVLINK_ESWITCH_MODE_SWITCHDEV:
> > + if (rswitch->mode == mode)
> > + return 0;
> > + rswitch->mode = mode;
> > + if (mode == DEVLINK_ESWITCH_MODE_SWITCHDEV)
> > + rvu_switch_enable(rvu);
> > + else
> > + rvu_switch_disable(rvu);
>
> I don't see the code handle creation and tearing down of representors.
>
> How do things work in this driver? Does AF have a representor netdev
> for each VF (that's separate from the VF netdev itself)? Those should
> only exist in switchdev mode, while legacy mode should use DMAC
> switching.
>
> I think what you want is a textbook VEPA vs VEB switch. Please take a
> look at drivers implementing .ndo_bridge_getlink/.ndo_bridge_setlink.
MCAM used for switching is present in AF and AF is not a netdev to
use ndo_bridge_setlink. Currently VF->VF/ PF->VF switching is possible
only with an external switch in hairpin mode (like VEPA), what we want to
achieve is switching based on DMAC in MCAM itself(internally).
We used DEVLINK_ESWITCH_MODE_SWITCHDEV as a trigger to
allocate MCAM rules. I understand now that it can be used to create
VF representors only. Can you please suggest any appropriate devlink
command we can use here so that new rules are created. Having these
rules by default will waste MCAM space.
Thanks,
Sundeep
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2021-07-20 17:58 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-07-19 8:59 [net-next PATCH 0/3] octeontx2-af: Introduce DMAC based switching Subbaraya Sundeep
2021-07-19 8:59 ` [net-next PATCH 1/3] octeontx2-af: Enable transmit side LBK link Subbaraya Sundeep
2021-07-19 8:59 ` [net-next PATCH 2/3] octeontx2-af: Prepare for allocating MCAM rules for AF Subbaraya Sundeep
2021-07-19 8:59 ` [net-next PATCH 3/3] octeontx2-af: Introduce internal packet switching Subbaraya Sundeep
2021-07-20 11:51 ` Jakub Kicinski
2021-07-20 17:56 ` sundeep subbaraya
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).